The ARM company is not just about the instruction set architecture. The ISA wouldn't be interesting at all if no good processors were built with the ISA [0]. For RISC-V to succeed, it requires a company that builds some good processor designs for it - for smartwatches, smartphones, tablets, laptops desktops - and licenses that to others. That company (one or multiple) does not (yet) exist, and is not easy to build.
[0]: Which is exactly why SPARC, and with one exception Power is dying, and why RISC V is yet to deliver. Nobody (bar IBM's POWER line) is building good processors with those ISAs that make it worth the effort to use. Nothing to do with the ISA - you just need chips people are interested in using.
Yep. It's difficult to build a community-- you need to have enough mass to get further interest in. From a business view you end up with the question of "Why bother with RISC-V when ARM is doing what we need and has enough critical mass to keep things going forward?"
About the only thing that could force that to change would be another company buying up ARM and changing the licensing mechanisms (e.g. pricing or even removing some license options) going forward.. or just wrecking the product utterly.
I do think RISC-V has an opportunity here, but only if ARM sells out to NV and NV screws this up as hard as they're likely to in that situation.
> For RISC-V to succeed, it requires a company that builds some good processor designs for it - for smartwatches, smartphones, tablets, laptops desktops - and licenses that to others
The way I see it is that this may actually generate incentive for someone to do that. One of the reasons that that isn't happening yet is because there's no real need with ARM vendors supplying and no real chance with ARM vendors as competition. This could, in theory, clear the way.
This is assuming it would have to be a new company, rather than an existing company like Qualcomm or AMD which could produce a processor with a different ISA if nVidia/ARM became unreasonable to deal with.
This is particularly true for Android because basically the entire thing is written in portable languages and the apps even run on a bytecode VM already, so switching to another architecture or even supporting multiple architectures at the same time wouldn't be that hard.
> This is particularly true for Android because basically the entire thing is written in portable languages and the apps even run on a bytecode VM already, so switching to another architecture or even supporting multiple architectures at the same time wouldn't be that hard.
Google could easily afford to design their own RISC-V CPUs and port Android to it, if they thought it was in their strategic interests to do so.
I think it really depends on how nVidia-owned Arm behaves. If it behaves the same as Softbank-owned Arm, I don't think Google would bother. If it starts to behave differently, in a way which upsets the Android ecosystem, Google might do something like this. (I imagine they'll give it some time to see whether Arm's behaviour changes post-acquisition.)
[0]: Which is exactly why SPARC, and with one exception Power is dying, and why RISC V is yet to deliver. Nobody (bar IBM's POWER line) is building good processors with those ISAs that make it worth the effort to use. Nothing to do with the ISA - you just need chips people are interested in using.